Normalization of RNA-sequencing data from samples with varying mRNA levels

Area of Science:

• Genomics
• Developmental Biology
• Bioinformatics

Background:

• RNA-sequencing (RNA-Seq) is a powerful tool for gene expression analysis.
• Standard normalization methods often assume equal total expression across samples.
• Global gene expression shifts can occur due to biological processes like development.

Purpose of the Study:

• To evaluate RNA-Seq normalization methods under conditions of fluctuating total RNA content.
• To compare the accuracy of Reads Per Kilobase Per Million (RPKM), Trimmed Mean of M-values (TMM), and Biological Scaling Normalization (BSN).
• To assess performance during zebrafish early developmental stages.

Main Methods:

• Utilized reverse transcription-quantitative PCR (RT-qPCR) as a benchmark for gene expression quantification.
• Applied RPKM, TMM, and BSN normalization techniques to RNA-Seq data from zebrafish embryos.
• Compared normalization method accuracy against RT-qPCR results.

Main Results:

• RPKM and TMM normalization methods produced systematically biased gene expression estimates.
• Biological Scaling Normalization (BSN) demonstrated improved accuracy in estimating transcript level dynamics.
• BSN's performance was superior when total RNA content varied significantly.

Conclusions:

• Traditional normalization methods (RPKM, TMM) are unreliable when total RNA content differs between samples.
• BSN is a more accurate normalization strategy for RNA-Seq studies with variable total RNA, such as developmental processes.
• Findings impact the interpretation of past and future RNA-Seq studies involving samples with differing RNA quantities.